Promotional Uplift Forecasting: Modelling the Lift Before You Commit the Discount

Promotional uplift forecasting answers that question before you commit the money. It estimates how many extra units a promotion will genuinely add, what it will cost, and what it leaves behind once the price goes back up.

Why the Sales Spike Is Not the Uplift

The number most teams report after a promotion is total sales during the offer period. That figure mixes several things together.

• Baseline sales. Units you would have sold at full price anyway. Discounting them is pure margin given away.

• True incremental sales. New shoppers, larger baskets and purchases that would not otherwise have happened. This is the real uplift.

• Pull forward. Loyal buyers stocking up early. These sales are borrowed from the weeks ahead and show up later as a dip.

• Cannibalisation. Shoppers switching from another of your own SKUs to the discounted one. Your total does not change, only the mix.

• Halo. Extra sales of related products because shoppers arrived for the deal. This is a genuine benefit worth counting.

A promotion that doubles sales during the offer can still lose money if most of those units were baseline, pull forward, or cannibalized.

Step 1: Establish a Clean Baseline

The baseline is your best estimate of what the SKU would sell at full price over the same period. Build it from recent non-promotional weeks, then adjust for seasonality, known events and trend. On marketplaces, exclude periods when the product was out of stock or lost the buy box, because those weeks understate normal demand.

If the baseline is wrong, every number that follows is wrong too. This is the step most worth getting right.

Step 2: Estimate the Lift by Discount Depth

Look at past promotions on the same SKU or similar SKUs in the same category. For each one, compare actual sales with the baseline at different discount depths. Most products show diminishing returns: the move from 10 to 20 percent off adds more than the move from 30 to 40 percent.

On Indian marketplaces, account for what shoppers actually see. A discount on its own behaves differently from one combined with a deal badge, a bank offer, or placement in a sale event. Tag each past promotion with these conditions, so the model compares like with like.

Step 3: Adjust for What Happens Around the Promotion

Estimate cannibalisation by checking what happened to sales of your closely related SKUs during past promotions. Estimate pull forward by measuring how far sales fell below baseline in the two to four weeks after. Estimate halo by tracking related products that were not discounted.

Competitor activity matters too. If a close competitor is likely to run a deeper discount in the same window, your lift will be smaller than history suggests.

Step 4: Do the Margin Arithmetic

With incremental units estimated, compare cost against gain.

• Discount cost on all units sold, including baseline units

• Any funding contribution to marketplace deal programs

• Additional advertising spend to support visibility

• Marketplace fees on the extra units

• Expected returns, which tend to rise with discount-led purchases

Against that, count the contribution from genuinely incremental units and halo sales, minus the margin lost in the post-promotion dip. If the result is negative, the promotion only makes sense for a stated strategic reason, such as acquiring new customers or clearing stock, and you should write that reason down.

Step 5: Check Stock Before You Commit

A promotion that sells out halfway through wastes the remaining ad spend and can cost search rank on the way back. Compare forecast units with available stock in the right fulfillment locations, including the buffer you need for the weeks after.

Learn From Every Promotion

After each promotion, compare the forecast with what happened. Record actual lift, the size of the post-sale dip, and the true profit. Over a few cycles, the model becomes specific to your brand and categories, and its forecasts become reliable enough to plan around.

Run small, controlled tests where you can. Promoting a SKU in one channel but not another, or varying discount depth across similar SKUs, gives clearer evidence than comparing this year with last year.

From Reaction to Planning

Most brands decide promotions under deadline pressure, when a marketplace deal invitation arrives or a competitor cuts price. Good promotional planning reverses that. It uses demand, pricing and availability data to forecast the lift in advance, so the team can decide which promotions to join, at what depth and on which SKUs, with the economics already on the table.

A discount is an investment. Forecasting the uplift is simply checking the return before you make it.